This paper proposes a learning method which employs an assistance for simplifying a motion to learn and learns it from easy to difficult gradually. Using Genetic Algorithm, it searches for the parameters of a controller appropriate for controlling the motion to learn with gradually increasing difficulty, i.e., with gradually decreasing degree of assistance. We show that this gradual search enables Genetic Algorithm to evolve a population of controllers efficiently by giving two examples: stable riding of a bicycle and stable control of a double inverted pendulum. A bicycle is much easier to control when it is running at a certain velocity. An initial velocity is given as assistance. The learning method searches for a controller gradually decreasing the initial velocity. Similarly a double inverted pendulum is much easier to control when an upward force supports the distal end of the pendulum. The learning method searches for a controller gradually decreasing the upward force. The reduction rate of assistance is adjustable in accordance with the adaptability of a population to the reduction.